Information processing method and device based on graph execution engine, equipment and medium

Abstract

The invention provides an information processing method and device based on a graph execution engine, equipment, a medium and a program product, and relates to the technical field of computers, in particular to the field of intelligent recommendation or information flow. According to the specific implementation scheme, in response to an information processing request, a plurality of target algorithm graphs are selected from an algorithm graph library; concurrently executing each target algorithm graph through a graph execution engine, and obtaining an information processing result returned after each target algorithm graph is executed; and summarizing the information processing results to obtain a response result of the information processing request. According to the technical scheme, decoupling of different information processing algorithms can be achieved, mutual influence of data and consumed time can be reduced in the concurrent execution process of the multiple information processing algorithms, and the computing resource utilization rate in the whole information processing process is increased.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, in particular to the field of intelligent recommendation or information flow, and specifically to an information processing method based on a graph execution engine, an information processing device based on a graph execution engine, an electronic device, a non-transitory computer-readable storage medium, and a computer program product. Background Art

[0002] In some information processing scenarios with complex business logic, in order to ensure the richness of information processing results, it is often necessary to try and combine a variety of information processing algorithms. For example, in complex information recommendation scenarios, it is often necessary to use a variety of different types of information recall algorithms.

[0003] In related technologies, in order to achieve unified acquisition of information processing results obtained by multiple information processing algorithms, it is often necessary to organize different information processing algorithms into a complete processing logic. Different information processing algorithms are developed by different development teams. When the processing logic is organized in a unified way, it often causes coupling in development, making the entire development process difficult and problems difficult to troubleshoot. Summary of the invention

[0004] The present disclosure provides an information processing method based on a graph execution engine, an information processing device based on a graph execution engine, an electronic device, a non-transitory computer-readable storage medium, and a computer program product.

[0005] According to one aspect of an embodiment of the present disclosure, there is provided an information processing method based on a graph execution engine, comprising:

[0006] In response to the information processing request, selecting a plurality of target algorithm graphs in the algorithm graph library;

[0007] Wherein, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing a set processing logic;

[0008] Concurrently executing each of the target algorithm graphs through a graph execution engine, and obtaining information processing results returned by each of the target algorithm graphs after execution is completed;

[0009] The information processing results are aggregated to obtain a response result to the information processing request.

[0010] According to another aspect of the embodiment of the present disclosure, there is also provided an information processing device based on a graph execution engine, including:

[0011] An algorithm graph selection module, for selecting a plurality of target algorithm graphs from an algorithm graph library in response to an information processing request;

[0012] Wherein, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing a set processing logic;

[0013] An algorithm graph concurrent execution module, used to concurrently execute each of the target algorithm graphs through a graph execution engine, and obtain information processing results returned by each of the target algorithm graphs after execution;

[0014] The result aggregation module is used to aggregate the information processing results to obtain a response result to the information processing request.

[0015] According to another aspect of the present disclosure, there is also provided an electronic device, including:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the information processing method based on the graph execution engine as described in any one of the embodiments of the present disclosure.

[0019] According to another aspect of an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is also provided, wherein the computer instructions are used to enable a computer to execute an information processing method based on a graph execution engine as described in any one of the embodiments of the present disclosure.

[0020] According to another aspect of an embodiment of the present disclosure, a computer program product is also provided, including a computer program, which, when executed by a processor, implements the steps of the information processing method based on the graph execution engine as described in any one of the embodiments of the present disclosure.

[0021] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0023] Figure 1 is a schematic diagram of an information processing method based on a graph execution engine provided according to an embodiment of the present disclosure;

[0024] Figure 2 is a schematic diagram of another information processing method based on a graph execution engine provided according to an embodiment of the present disclosure;

[0025] Figure 3 is a schematic diagram of another information processing method based on a graph execution engine provided according to an embodiment of the present disclosure;

[0026] Figure 4 It is a schematic diagram of the flow of a calculation result of an operator applicable to an embodiment of the present disclosure in a corresponding algorithm diagram or in other algorithm diagrams;

[0027] Figure 5 is a schematic diagram of another information processing method based on a graph execution engine provided according to an embodiment of the present disclosure;

[0028] Figure 6 is a scenario diagram of a specific business scenario that can implement the embodiments of the present disclosure;

[0029] Figure 7 is a structural diagram of an information processing device based on a graph execution engine provided according to an embodiment of the present disclosure;

[0030] Figure 8 It is a block diagram of an electronic device used to implement an information processing method based on a graph execution engine according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0031] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0032] Figure 1 It is a schematic diagram of an information processing method based on a graph execution engine provided according to an embodiment of the present disclosure. The embodiment of the present disclosure can be applied to the situation where, according to an information processing request, a graph execution engine is called to execute an algorithm graph of multiple information processing algorithms, and multiple types of information processing results are obtained for unified summary. The method can be executed by an information processing device based on a graph execution engine, which can be implemented in hardware and / or software, and can generally be configured in a server installed with a graph execution engine, typically, an information processing server.

[0033] Correspondingly, such as Figure 1 As shown, the method may specifically include:

[0034] S110. In response to the information processing request, select multiple target algorithm graphs from the algorithm graph library.

[0035] Among them, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing the set processing logic.

[0036] In this embodiment, a graph execution engine is used to execute the business logic of the information processing algorithm. The graph execution engine is mainly used to execute queries in the graph database and process graph data. It describes the business logic of the task to be executed as a DAG (Directed Acyclic Graph) by parsing the graph description language, and executes these tasks according to predetermined rules.

[0037] Among them, the various embodiments of the present disclosure are mainly used in complex business scenarios, that is, information processing scenarios with numerous information processing algorithms. At this time, multiple development teams independently develop independent algorithm graphs. In the various embodiments of the present disclosure, each development team is required to independently maintain the algorithm graph of the developed information processing algorithm. The information processing server can uniformly manage the above-mentioned algorithm graphs and store them together in the algorithm graph library. By having different development teams independently maintain independent algorithm graphs, development decoupling is achieved, which effectively reduces the difficulty of development. When abnormal problems occur, the development team to which the problematic algorithm graph belongs can be quickly located.

[0038] Accordingly, each algorithm graph can be understood as a DAG for implementing the set information processing function. Figure 1 Generally, it consists of nodes and directed edges. One node corresponds to an operator used to execute the set processing logic, for example, database operation operators such as association operators and merge operators. Directed edges are used to describe the execution order between operators. For example, for a directed edge "A->B" from operator A to operator B, it means that after the algorithm graph where "A->B" is located is triggered to execute, operator A is executed first, and after operator A is executed, operator B is executed.

[0039] Among them, the information processing request can be understood as a request for triggering the execution of one or more algorithm diagrams to perform information processing operations. The information processing request can be generated by different triggering methods in different application scenarios. For example, taking the information processing algorithm as a recall algorithm as an example, in the recommendation scenario of the feed stream, the information processing request can be triggered after the user swipes up or pulls down the recommendation page. Alternatively, in the user's active search scenario, the information processing request can be triggered after the user enters the set search keyword through the search input box, or, in the user's shopping scenario, the information processing request can be triggered when the user opens the purchase page of the set category.

[0040] Furthermore, the method of selecting multiple target algorithm graphs in the algorithm graph library may include: selecting all algorithm graphs in the algorithm graph library as target algorithm graphs, or randomly selecting a set number of algorithm graphs in the algorithm graph library as target algorithm graphs, or, based on one or more description fields contained in the information processing request, selecting multiple target algorithm graphs that match the description fields in the algorithm graph library, etc. This embodiment does not limit this.

[0041] Among them, the target algorithm diagram can be understood as an algorithm diagram used to obtain the required information processing results for this information processing request.

[0042] S120, executing each of the target algorithm graphs concurrently through a graph execution engine, and obtaining information processing results returned by each of the target algorithm graphs after execution is completed.

[0043] In this embodiment, after obtaining the above-mentioned target algorithm graphs, the execution of each target algorithm graph can be concurrently controlled based on the processing mechanism built into the graph execution engine. Accordingly, after each target algorithm graph is triggered to execute, based on the internally defined graph logic, each operator is triggered to execute the corresponding processing logic in sequence.

[0044] Among them, after each target algorithm graph is executed, it will return the information processing results obtained based on the information processing algorithm defined in the target algorithm graph.

[0045] In this embodiment, different target algorithm graphs may or may not have data dependencies during execution. If there is no data dependency, each target algorithm graph can be truly concurrently executed. If there is a data dependency, a common storage space can be pre-planned, and the dependent data can be stored in the common storage space to achieve sharing of the dependent data.

[0046] S130: Summarize the information processing results to obtain a response result to the information processing request.

[0047] Aggregating the information processing results can be understood as bringing together the information processing results. For example, after the target algorithm graph A is executed, 5 information processing results are obtained, and after the target calculation graph B is executed, 7 information processing results are obtained. By aggregating, a total of 5+7 information processing results are obtained.

[0048] In this embodiment, in addition to directly performing the above-mentioned aggregation process, after obtaining the information processing results returned by each target algorithm graph after execution, it is also possible to first perform simple data deduplication processing on each information processing result, and then aggregate the data of each information processing result after deduplication processing. Alternatively, it is also possible to perform certain screening operations on each information processing result after deduplication processing through the set post-processing logic, and then aggregate the data.

[0049] After obtaining the above data summary results, the summary results can be provided to the upstream service module of the information processing service, such as the information recommendation server, and the upstream service module performs subsequent processing on the summary results, such as finely sorting the summary results and making user recommendations.

[0050] The technical solution of the disclosed embodiment generates multiple independent algorithm graphs for different information processing algorithms, and when the actual information processing needs are detected, calls the graph execution engine to independently and concurrently trigger different algorithm graphs to obtain information processing results from different dimensions and feed them back to the upstream functional modules with information processing needs. It creatively proposes an information processing algorithm based on a multi-algorithm graph, which can decouple different information processing algorithms in complex scenarios where large-scale processing logic composed of multiple information processing algorithms is required for information processing, so as to quickly locate the problematic information processing algorithm when abnormal problems occur. At the same time, in the concurrent execution of multiple information processing algorithms after decoupling, the mutual influence of data and time consumption can be reduced, thereby improving the utilization rate of computing resources in the entire information processing process.

[0051] Figure 2 It is a schematic diagram of another information processing method based on a graph execution engine provided according to an embodiment of the present disclosure. This embodiment is refined based on the above embodiments. In this embodiment, the operation of "selecting multiple target algorithm graphs in an algorithm graph library in response to an information processing request" is specifically concretized.

[0052] Correspondingly, such as Figure 2 As shown, the method may specifically include:

[0053] S210. In response to an information processing request, extract at least one key information description field in the information processing request.

[0054] The key information description field may be understood as a field that describes the request details of the information processing request. For example, the key information description field may include a request scenario description field, such as a feed stream recommendation scenario field, a search scenario field, or a shopping homepage recommendation field, etc. Alternatively, the key information description field may include a requester attribute description field, such as a requester basic information field (such as age, gender, or occupation, etc.), or a current location field, etc.

[0055] Of course, it is understandable that the acquisition of the above-mentioned requester attribute description field needs to be authorized in advance by the requester, and can be obtained after the requester completes the authorization.

[0056] Of course, it is understandable that the key information description field may also include an information processing accuracy description field and an experiment description field, and this embodiment does not limit this. The information processing accuracy description field is used to describe the accuracy requirements for the current information processing results, for example, it may be 80% or 90%. The experiment description field is used to describe whether the current information processing process is a set experiment process or a normal information processing process.

[0057] Specifically, at least one key information description field extracted from the information processing request is used to select target algorithm graphs for implementing the current information processing from the algorithm graph library.

[0058] S220. According to the key information description field, query the mapping relationship between the pre-constructed description information and the algorithm graph, and select each of the target algorithm graphs in the algorithm graph library.

[0059] In this embodiment, different types of mapping relationships can be pre-established to describe the target algorithm graphs required to be selected under different types of description information. Accordingly, after obtaining one or more key information description fields contained in the information processing request, a query can be performed in one or more pre-established mapping relationships based on the key information description fields to obtain the target algorithm graph hit by the query.

[0060] In an optional implementation of this embodiment, the key information description field may include: a requester attribute description field and a request scenario description field.

[0061] Correspondingly, according to the key information description field, querying the mapping relationship between the pre-constructed description information and the algorithm graph, and selecting each of the target algorithm graphs in the algorithm graph library may specifically include:

[0062] According to the requester attribute description field, query a first mapping relationship between the requester attribute information and the algorithm graph; according to the first mapping relationship, select a first type of algorithm graph in the algorithm graph library;

[0063] According to the request scene description field, query the second mapping relationship between the request scene information and the algorithm graph; according to the second mapping relationship, select the second type of algorithm graph in the algorithm graph library;

[0064] According to the first type of algorithm graph and the second type of algorithm graph, each of the target algorithm graphs is obtained.

[0065] In a specific example, the requester attribute description field can be used to include the region information of the requester, and then, a pre-established form such as: "Region 1: Algorithm Figure 1 ,algorithm Figure 2 ,algorithm Figure 5 ; Region 2: Algorithms Figure 3 ,algorithm Figure 4 ", and then, after obtaining the specific geographical information in the requester attribute description field, the first type of algorithm graph corresponding to the requester attribute description field can be determined in the first mapping relationship. Similarly, the second type of algorithm graph that is adapted can be determined based on the request scenario description field. Furthermore, by combining the above-mentioned first type of algorithm graph and the second type of algorithm graph, a complete target algorithm graph can be obtained.

[0066] Through the above settings, the applicable scenarios of information processing services can be further enriched, and users' different types of information processing needs can be customized to meet them, thereby further improving the hit rate of information processing results for users' actual needs.

[0067] Based on the above embodiments, the key information description field may further include: an information processing accuracy description field.

[0068] Accordingly, obtaining each of the target algorithm graphs according to the first type of algorithm graph and the second type of algorithm graph may further include:

[0069] Calling the information processing accuracy determination model to obtain the information processing accuracy of each of the first-type algorithm graphs for the requester attribute description field, and the information processing accuracy of each of the second-type algorithm graphs for the request scenario description field;

[0070] Among the first type of algorithm diagrams and the second type of algorithm diagrams, select the algorithm diagrams whose information processing accuracy is higher than the information processing accuracy description field;

[0071] The target algorithm graphs are obtained according to the filtered algorithm graphs.

[0072] In this embodiment, an information processing accuracy description field may be further carried in the information processing request, in which the information processing accuracy required for this information processing process is defined. Based on the above information processing accuracy description field, the first type of algorithm graph and the second type of algorithm graph screened based on the requester attribute description field and the request scenario description field need to be screened again to retain the algorithm graph that meets the information processing accuracy requirements, and finally form the target algorithm graph.

[0073] Generally speaking, different types of algorithm graphs have different information processing accuracy for different types of requester attribute information or different request scenario information. For example, algorithm graph A has a higher information processing accuracy for the "feed stream recommendation scenario", while the information processing accuracy for the "information search scenario" is slightly worse, but it is also usable. If the user does not have actual information processing accuracy requirements, algorithm graph A can be used as the target algorithm graph to return the information processing results under the above-mentioned types of request scenario description fields. However, when the user has a higher requirement for information processing accuracy, the algorithm graph A will be discarded in the "information search scenario".

[0074] In an optional implementation of this embodiment, an information processing accuracy determination model can be pre-constructed, and the information processing accuracy determination model is used to evaluate the accuracy of the information processing results obtained by processing the set algorithm diagram based on the descriptive information defined in the key information description field of the set type.

[0075] Through the above settings, the applicable scenarios of information processing services can be further enriched, and users' different types of information processing needs can be customized to meet them, thereby further improving the hit rate of information processing results for users' actual needs.

[0076] In another optional implementation of this embodiment, the key information description field includes: an experiment description field;

[0077] Accordingly, the querying of the mapping relationship between the pre-built description information and the algorithm graph according to the key information description field and the selection of each target algorithm graph in the algorithm graph library may include:

[0078] According to the experiment description field, query the third mapping relationship between the experiment information and the algorithm diagram;

[0079] According to the third mapping relationship, a third type of algorithm graph is selected from the algorithm graph library as each of the target algorithm graphs.

[0080] In this embodiment, an experiment description field may be further carried in the information processing request, and the experiment description field is used to describe whether the current information processing request is an actual request triggered by a normal user and received by the information processing server when it is operating normally online, or an experimental request simulated and generated by R&D personnel during the version iteration process of setting the algorithm diagram.

[0081] If it is determined that the information processing request belongs to a set experiment based on the experiment description information defined in the experiment description field, it is necessary to obtain multiple specific third-category algorithm diagrams required to be verified in the experiment from the pre-established third mapping relationship as the target algorithm diagram for executing the information processing service this time.

[0082] Through the above settings, the applicable scenarios of information processing services can be further enriched, and users' different types of information processing needs can be customized to meet them, thereby further improving the hit rate of information processing results for users' actual needs.

[0083] S230, executing each of the target algorithm graphs concurrently through a graph execution engine, and obtaining information processing results returned by each of the target algorithm graphs after execution is completed.

[0084] S240: Summarize the information processing results to obtain a response result to the information processing request.

[0085] The technical solution of the disclosed embodiment extracts at least one key information description field in the information processing request in response to the information processing request, selects the implementation method of each target algorithm graph in the algorithm graph library according to the mapping relationship between the pre-built description information and the algorithm graph, and can aggregate all independent algorithm graphs required for information processing in the graph execution engine for unified management, and then can selectively determine the target algorithm graph that best matches the information processing request according to the specific characteristics of the information processing request received in real time, and complete the information processing operation of the information. It realizes the integrated satisfaction of different information processing requirements and reduces the complexity of the information processing server.

[0086] Figure 3 It is a schematic diagram of another information processing method based on a graph execution engine provided according to an embodiment of the present disclosure. This embodiment is refined based on the above embodiments. In this embodiment, the operation of "concurrently executing each of the target algorithm graphs through a graph execution engine" is specifically concretized.

[0087] Correspondingly, such as Figure 3 As shown, the method may specifically include:

[0088] S310. In response to an information processing request, select multiple target algorithm graphs from an algorithm graph library.

[0089] Among them, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing the set processing logic.

[0090] S320, respectively obtain the data type of the pre-dependency data required by each of the target algorithm graphs when executing.

[0091] In this embodiment, it is considered that when the algorithm graph completes information processing, it needs to use certain pre-dependency data as the overall input of the algorithm graph (typically, the input of the first operator in the algorithm graph), or the input required by the non-first operator in the algorithm graph. Therefore, before triggering the execution of each target algorithm graph, it is necessary to first determine the data type of the pre-dependency data specifically required by each target algorithm graph when executing.

[0092] The pre-dependency data required by an algorithm graph may be directly obtained, or may need to be output by other algorithm graphs during execution or after execution. This embodiment does not limit this.

[0093] In a specific example, during the execution of algorithm graph A, it is necessary to obtain the attribute information of the requester (for example, the time when the request is sent, the geographical information of the requester, or the age information of the requester, etc.); during the execution of algorithm graph B, it is necessary to obtain the requester's most recent shopping behavior data; and during the execution of algorithm graph C, it is necessary to obtain the information processing results output by algorithm graph A, for example, it is necessary to filter out the hot events that the requester may be interested in from all the hot events processed by algorithm graph A.

[0094] It is understandable that the pre-dependency data of the data types required by algorithm graph A and algorithm graph B can generally be directly obtained, while the pre-dependency data of the data type required by algorithm graph C can only be obtained after algorithm graph A is executed.

[0095] S330. According to the data type and the information processing request, execute an operation of acquiring all the pre-dependency data required for each target algorithm graph.

[0096] Specifically, the pre-dependent data of the set data type can be directly obtained in the information processing request, for example, the requester scenario information contained in the requester scenario description field, or the requester attribute information contained in the requester attribute description field, etc., or the pre-dependent data of the set data type can also be obtained from other databases in combination with the requester identification information in the information processing request, for example, the historical behavior data of the requester, etc. Of course, there are also pre-dependent data of the set data type that cannot be directly obtained before the concurrent execution of each target algorithm graph.

[0097] Therefore, in combination with the data type and the information processing request, an operation of obtaining all the prerequisite dependency data required for each target algorithm graph can be performed, and each prerequisite dependency data that can be obtained can be obtained as much as possible for direct use by subsequent target algorithm graphs.

[0098] S340: construct a common precondition graph according to the acquired precondition dependency data.

[0099] In this embodiment, after all available pre-dependency data are acquired, a public pre-dependency graph may be constructed based on the acquired pre-dependency data. The process of constructing the public pre-dependency graph may be understood as a unified acquisition process of public data dependencies.

[0100] It is understandable that, since each embodiment of the present disclosure independently constructs different algorithm graphs corresponding to different information processing algorithms, each algorithm Figure 1 Generally, it is necessary to independently obtain the required pre-dependent data. For example, if there are three algorithm graphs that all need to use the requester scene information, then these three algorithm graphs will request three times in total to obtain the same requester scene information. This operation brings a large amount of repeated data calculations, reducing the computational efficiency of the information processing process. In this embodiment, in order to avoid the above-mentioned problems caused by the concurrent use of multiple independent algorithm graphs, it is proposed to uniformly configure all pre-dependent data in a common pre-dependent graph to avoid repeated acquisition of common pre-dependent data.

[0101] S350: After using the common pre-graph as the common input graph of each of the target algorithm graphs, the target algorithm graphs are concurrently executed by a graph execution engine.

[0102] By setting the common predecessor graph as the common input graph of each of the target algorithm graphs, each of the target algorithm graphs can directly obtain the required predecessor dependency data from the common input graph after being triggered for execution.

[0103] In an optional implementation of this embodiment, after the target algorithm graphs are concurrently executed by the graph execution engine, the following may also be included:

[0104] After each operator in each of the target algorithm graphs is triggered to execute, the matching processing logic is executed according to the calculation results obtained from the preceding operators of the target algorithm graph to which it belongs and / or the preceding dependency data obtained from the common preceding graph;

[0105] When it is determined that a target calculation result obtained by executing any operator is relied upon by operators of other algorithm graphs, the target calculation result is updated in real time to the common pre-graph.

[0106] In this optional implementation, each operator in the target algorithm graph is specifically taken as the observation object to describe the execution of each operator. It is understandable that after each target algorithm graph is triggered and executed concurrently, each operator in each target algorithm graph can be executed sequentially according to its internally defined execution logic (the connection method of different directed edges to each node). In general, when each operator is executed, it has to use the set input data. There are two main sources of the above input data, one is the calculation result output by the operator that completes the calculation in the previous order in the same target algorithm graph, and the other is the pre-dependency data obtained from the public pre-graph. The pre-dependency data that needs to be obtained from the public pre-graph is divided into two cases. One is that each target algorithm graph is written into the public pre-graph before being triggered and executed concurrently, and the other is that an operator in other target algorithm graphs is written into the public pre-graph after execution.

[0107] Specifically, in Figure 4 A schematic diagram showing the flow of a calculation result of an operator applicable to an embodiment of the present disclosure in the corresponding algorithm diagram or in other algorithm diagrams is shown in FIG.

[0108] like Figure 4 As shown, assuming that for information processing requests, a total of algorithms are triggered Figure 1 ,algorithm Figure 2 And the algorithm Figure 3 A total of 3 algorithm graphs are executed concurrently. Among them, Figure 4 The node "V" in represents an operator, "d" represents an operator calculation result generated in an independent algorithm graph and only circulated and used in the independent algorithm graph (can be called a local calculation result). "D1" and "D2" represent operator calculation results generated in an independent algorithm graph but need to be used by operators in other algorithm graphs (can be called global calculation results). It can be understood that if a global calculation result needs to be written into the common front graph.

[0109] Through the above settings, the dependencies between different algorithm graphs can be eliminated to the greatest extent, and the complex interaction logic between algorithm graphs can be avoided, which reduces the mutual influence of different algorithm graphs on data from the source and improves development efficiency.

[0110] On the basis of the above embodiments, after each operator in each of the target algorithm graphs is triggered to execute, the matching processing logic is executed according to the calculation results obtained from the preceding operators of the target algorithm graph to which it belongs and / or the preceding dependency data obtained from the common preceding graph, which may specifically include:

[0111] After the target operator is triggered to execute, detecting whether the target operator needs to obtain target predecessor dependency data from the common predecessor graph during execution;

[0112] When it is determined that the target predecessor dependency data needs to be obtained from the public predecessor graph, detecting whether the target predecessor dependency data is currently stored in the public predecessor graph;

[0113] If the target pre-dependency data is stored, executing the processing logic matching the target operator according to the target pre-dependency data;

[0114] If the target pre-dependency data is not stored, then the operation of detecting whether the target pre-dependency data is currently stored in the public pre-dependency graph is returned to be executed at every preset time interval until the target pre-dependency data is detected.

[0115] In this embodiment, after each operator is triggered to execute, it first needs to detect whether the input data it needs to use is ready. If the data is ready, the internal processing logic can be implemented directly. If the data is not ready, it is necessary to wait until the data is ready before implementing the corresponding processing logic.

[0116] Specifically, after the target operator is triggered to execute, it first detects whether it needs to obtain the target pre-dependency data from the public pre-graph. If not, it means that the target pre-dependency data required by the target operator comes from the calculation results of other pre-order operators in the target algorithm graph where the target operator is located. At this time, the processing logic matching the target operator can be directly executed.

[0117] When the target operator needs to obtain the target pre-dependency data from the public pre-dependency graph during execution, and the target pre-dependency data is not temporarily stored in the public pre-dependency graph, the target operator can suspend execution first, and query the target pre-dependency data in the public pre-dependency graph at regular intervals. Once the target pre-dependency data is generated and stored in the public pre-dependency graph, the target operator can immediately implement the corresponding processing logic.

[0118] Through the above settings, each operator in each target algorithm graph only needs to communicate with the public predecessor graph when executing, and there is no need to establish a logical relationship with other operators in other target algorithm graphs. This maximizes the logical decoupling between different algorithm graphs and greatly improves the code development efficiency.

[0119] S360, obtaining the information processing results returned by each of the target algorithm graphs after execution.

[0120] S370: Summarize the information processing results to obtain a response result to the information processing request.

[0121] The technical solution of the disclosed embodiment obtains the data type of the prerequisite dependency data required by each target algorithm graph when executing; performs the acquisition operation of all prerequisite dependency data required by each target algorithm graph according to the data type and the information processing request; constructs a common prerequisite graph according to the acquired prerequisite dependency data; and uses the common prerequisite graph as the common input graph of each target algorithm graph, and then concurrently executes each target algorithm graph through a graph execution engine, thereby achieving decoupling between different algorithm graphs. Even in complex information processing scenarios, each algorithm graph only needs to exchange data with the common prerequisite graph, thus avoiding the complex data exchange process between different algorithm graphs, reducing the mutual influence between different operator graphs in data and time consumption, and improving the efficiency of information processing.

[0122] On the basis of the above embodiments, the summarizing of the information processing results may include:

[0123] After each of the information processing results is post-processed according to the data processing logic defined in the pre-constructed public post-processing graph, the post-processing results are summarized.

[0124] In this embodiment, a public post-processing graph is further constructed for post-processing each information processing result. Based on the public post-processing graph, the configured data processing logic (e.g., data filtering logic) can be performed on each of the information processing results returned by each target algorithm graph to meet the required information processing requirements.

[0125] Through the above settings, a unified data post-processing process can be set for each information processing result returned by each target algorithm graph to meet different information processing requirements in different scenarios.

[0126] Figure 5 It is a schematic diagram of another information processing method based on a graph execution engine provided according to an embodiment of the present disclosure. This embodiment is refined based on the above embodiments. In this embodiment, the operation of "post-processing each of the information processing results according to the data processing logic defined in the pre-built public post-graph, and aggregating the post-processing results" is concretized.

[0127] Correspondingly, such as Figure 5 As shown, the method may specifically include:

[0128] S510. In response to an information processing request, select multiple target algorithm graphs from an algorithm graph library.

[0129] Among them, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing the set processing logic.

[0130] S520, executing each of the target algorithm graphs concurrently through a graph execution engine, and obtaining information processing results returned by each of the target algorithm graphs after execution is completed.

[0131] S530: Count the total number of information result entries included in each of the information processing results.

[0132] In this embodiment, it is defined in the common post-graph that: when the total number of entries of each information processing result returned by each target algorithm graph is too large, the above-mentioned information processing results are deleted and filtered according to the importance of each target algorithm graph, so that the total number of remaining information processing results meets the quantity requirement of the upstream service module for the information processing results.

[0133] Among them, the number of information processing results returned by each target algorithm graph after execution can be obtained, and the above-mentioned numbers are accumulated and summed up to obtain the total number of information result entries.

[0134] S540, determine whether the total number of information result items exceeds a preset quantity threshold: if so, execute S550; otherwise, execute S560.

[0135] If the total number of information result entries exceeds the preset number threshold, it is necessary to delete and filter each information processing result before summarizing the results; if the total number of information result entries does not exceed the preset number threshold, it is not necessary to delete and filter, and the information processing results are directly summarized.

[0136] S550. Calculate the difference between the total number of information result entries and the quantity threshold as the number of deleted entries, and execute S570.

[0137] S560. Directly summarize the information processing results and execute S5100.

[0138] S570, respectively obtain the importance score of each of the target algorithm graphs, and determine the deletion ratio of the result items of each of the target algorithm graphs according to the importance score, and execute S580.

[0139] In this embodiment, an importance evaluation model can be pre-built, which is based on prior evaluation knowledge of different algorithm graphs, or trained by recommending user satisfaction scores of information processing results of different algorithm graphs. The importance evaluation model can be updated in real time.

[0140] Among them, by inputting each of the target algorithm graphs into the importance evaluation model respectively, the importance score of each target algorithm graph can be obtained.

[0141] Based on the importance score, the deletion ratio of the result items of each target algorithm graph can be determined. The deletion ratio of the result items can be understood as the proportion of information processing results that need to be deleted among all the information processing results returned by a target algorithm graph. In a specific example, the higher the importance score of a target algorithm graph, the smaller the deletion ratio of its result items.

[0142] S580, after sorting the information processing results of each of the target algorithm graphs, deletion and filtering are performed according to the deletion ratio of the matching result items.

[0143] S590. Summarize the deletion and filtering results corresponding to each of the target algorithm graphs and execute S5100.

[0144] S5100. Obtain a response result to the information processing request based on the summary result.

[0145] The technical solution of the disclosed embodiment can further enrich the applicable scenarios of information processing services, improve the matching degree between information processing results and actual information processing needs, and improve the accuracy of information processing results fed back to upstream functional modules, thereby improving the hit rate of information processing results for actual user needs and further improving user experience, by performing post-processing on each of the information processing results based on the data processing logic defined in the common post-processing graph.

[0146] Figure 6 is a scene diagram of a specific business scenario that can implement the embodiment of the present disclosure. Figure 6 As shown, the box shows a schematic diagram of the complete process of the information processing server generating corresponding summary results for the input information processing request.

[0147] Specifically, when obtaining the information processing request, a preset selection algorithm can be used to first determine which algorithm graphs will be selected for use for this information processing service (that is, the target algorithm graph). Then, based on the selected target algorithm graph and the information processing request, a common premise graph for producing premise dependency data is first constructed. The common premise graph can communicate with each target algorithm graph separately to enable the operators in each target algorithm graph to obtain input data from the common premise graph to execute calculation logic, or to update the calculated data and write it to the common premise graph.

[0148] After completing the configuration of the public pre-graph, the above-mentioned target algorithm graphs can be triggered concurrently to return (or output) the corresponding information processing results. Afterwards, the data processing post-process is executed through the data processing logic defined in the public post-graph to delete and filter each information processing result, and the information summary result obtained after deletion and filtering is fed back.

[0149] As an implementation of the above-mentioned information processing methods based on the graph execution engine, the present disclosure also provides an optional embodiment of an execution device for implementing the above-mentioned information processing methods based on the graph execution engine.

[0150] Figure 7 is a structural diagram of an information processing device based on a graph execution engine provided according to an embodiment of the present disclosure. Figure 7 As shown, the device includes: an algorithm graph selection module 710, an algorithm graph concurrent execution module 720 and a result summary module 730, wherein:

[0151] The algorithm graph selection module 710 is used to select a plurality of target algorithm graphs from the algorithm graph library in response to the information processing request;

[0152] Wherein, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing a set processing logic;

[0153] The algorithm graph concurrent execution module 720 is used to concurrently execute each of the target algorithm graphs through a graph execution engine, and obtain information processing results returned by each of the target algorithm graphs after execution;

[0154] The result aggregation module 730 is used to aggregate the information processing results and obtain a response result to the information processing request.

[0155] The technical solution of the disclosed embodiment generates multiple independent algorithm graphs for different information processing algorithms, and when the actual information processing needs are detected, calls the graph execution engine to independently and concurrently trigger different algorithm graphs to obtain information processing results from different dimensions and feed them back to the upstream functional modules with information processing needs. It creatively proposes an information processing algorithm based on a multi-algorithm graph, which can decouple different information processing algorithms in complex scenarios where large-scale processing logic composed of multiple information processing algorithms is required for information processing, so as to quickly locate the problematic information processing algorithm when abnormal problems occur. At the same time, in the concurrent execution of multiple information processing algorithms after decoupling, the mutual influence of data and time consumption can be reduced, thereby improving the utilization rate of computing resources in the entire information processing process.

[0156] Based on the above embodiments, the algorithm diagram selection module may include:

[0157] a key information description field extraction unit, configured to extract at least one key information description field in the information processing request in response to the information processing request;

[0158] A mapping selection unit is used to query the mapping relationship between the pre-constructed description information and the algorithm diagram according to the key information description field, and select each of the target algorithm diagrams in the algorithm diagram library.

[0159] On the basis of the above embodiments, the key information description field may include: a requester attribute description field and a request scenario description field;

[0160] Accordingly, the mapping selection unit may include:

[0161] A first selection subunit is used to query a first mapping relationship between the requester attribute information and the algorithm graph according to the requester attribute description field; and select a first type of algorithm graph in the algorithm graph library according to the first mapping relationship;

[0162] A second selection subunit is used to query a second mapping relationship between the request scene information and the algorithm graph according to the request scene description field; and select a second type of algorithm graph in the algorithm graph library according to the second mapping relationship;

[0163] A result summary subunit is selected to obtain each of the target algorithm graphs based on the first type of algorithm graph and the second type of algorithm graph.

[0164] On the basis of the above embodiments, the key information description field may further include: an information processing accuracy description field;

[0165] The selection result summary subunit can be specifically used for:

[0166] Calling the information processing accuracy determination model to obtain the information processing accuracy of each of the first-type algorithm graphs for the requester attribute description field, and the information processing accuracy of each of the second-type algorithm graphs for the request scenario description field;

[0167] Among the first type of algorithm diagrams and the second type of algorithm diagrams, select the algorithm diagrams whose information processing accuracy is higher than the information processing accuracy description field;

[0168] The target algorithm graphs are obtained according to the filtered algorithm graphs.

[0169] Based on the above embodiments, the key information description field may include: an experiment description field;

[0170] The mapping selection unit may be used to:

[0171] According to the experiment description field, query the third mapping relationship between the experiment information and the algorithm diagram;

[0172] According to the third mapping relationship, a third type of algorithm graph is selected from the algorithm graph library as each of the target algorithm graphs.

[0173] Based on the above embodiments, the algorithm graph concurrent execution module can be used to:

[0174] Respectively obtain the data type of the pre-dependency data required by each target algorithm graph when executing;

[0175] According to the data type and the information processing request, executing an acquisition operation of all the pre-dependent data required for each of the target algorithm graphs;

[0176] Constructing a common precondition graph according to the acquired precondition dependency data;

[0177] After the common pre-graph is used as the common input graph of each of the target algorithm graphs, each of the target algorithm graphs is executed concurrently through a graph execution engine.

[0178] Based on the above embodiments, an operator execution module may be further included, which is used to:

[0179] After the target algorithm graphs are concurrently executed by the graph execution engine, each operator in each target algorithm graph, after being triggered to execute, executes matching processing logic according to the calculation results obtained from the predecessor operator of the target algorithm graph to which it belongs and / or the predecessor dependency data obtained from the common predecessor graph;

[0180] When it is determined that a target calculation result obtained by executing any operator is relied upon by operators of other algorithm graphs, the target calculation result is updated in real time to the common pre-graph.

[0181] Based on the above embodiments, the operator execution module may be further used for:

[0182] After the target operator is triggered to execute, detecting whether the target operator needs to obtain target predecessor dependency data from the common predecessor graph during execution;

[0183] When it is determined that the target predecessor dependency data needs to be obtained from the public predecessor graph, detecting whether the target predecessor dependency data is currently stored in the public predecessor graph;

[0184] If the target pre-dependency data is stored, executing the processing logic matching the target operator according to the target pre-dependency data;

[0185] If the target pre-dependency data is not stored, then the operation of detecting whether the target pre-dependency data is currently stored in the public pre-dependency graph is returned to be executed at every preset time interval until the target pre-dependency data is detected.

[0186] Based on the above embodiments, the result aggregation module can be used to:

[0187] After each of the information processing results is post-processed according to the data processing logic defined in the pre-constructed public post-processing graph, the post-processing results are summarized.

[0188] Based on the above embodiments, the result summary module can be further used to:

[0189] Counting the total number of information result entries contained in each of the information processing results;

[0190] If the total number of information result entries exceeds a preset number threshold, the difference between the total number of information result entries and the number threshold is calculated as the number of deleted entries;

[0191] Obtaining the importance scores of each of the target algorithm graphs respectively, and determining the deletion ratio of the result items of each of the target algorithm graphs according to the importance scores;

[0192] After sorting the information processing results of each of the target algorithm graphs, deletion and filtering are performed according to the deletion ratio of the matching result items;

[0193] The deletion and filtering results corresponding to each of the target algorithm graphs are summarized.

[0194] The above-mentioned product can execute the method provided by any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method.

[0195] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0196] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0197] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0198] like Figure 8 As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0199] A number of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0200] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as performing the information processing method based on the graph execution engine as described in any of the embodiments of the present disclosure. For example, in some embodiments, the information processing method based on the graph execution engine as described in any of the embodiments of the present disclosure may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the information processing method based on the graph execution engine as described in any of the embodiments of the present disclosure described above may be executed. Alternatively, in other embodiments, the computing unit 801 may be configured to execute the information processing method based on the graph execution engine as described in any of the embodiments of the present disclosure in any other appropriate manner (for example, by means of firmware). That is:

[0201] In response to the information processing request, selecting a plurality of target algorithm graphs in the algorithm graph library;

[0202] Wherein, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing a set processing logic;

[0203] Concurrently executing each of the target algorithm graphs through a graph execution engine, and obtaining information processing results returned by each of the target algorithm graphs after execution is completed;

[0204] The information processing results are aggregated to obtain a response result to the information processing request.

[0205] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0206] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0207] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0208] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0209] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0210] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services. The server may also be a server of a distributed system, or a server combined with a blockchain.

[0211] Artificial intelligence is a discipline that studies how to use computers to simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.). It includes both hardware-level and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural language processing technology, machine learning / deep learning technology, big data processing technology, knowledge graph technology, and other major directions.

[0212] Cloud computing refers to a technology system that uses network access to elastically scalable shared physical or virtual resource pools. Resources can include servers, operating systems, networks, software, applications, and storage devices, and can be deployed and managed on demand and in a self-service manner. Cloud computing technology can provide efficient and powerful data processing capabilities for technical applications such as artificial intelligence and blockchain, as well as model training.

[0213] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions provided by this disclosure can be achieved, and this document does not limit this.

[0214] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. An information processing method based on a graph execution engine, comprising: In response to the information processing request, multiple target algorithm graphs are selected from the algorithm graph library; wherein one algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph includes at least one operator for executing a set processing logic; Concurrently executing each of the target algorithm graphs through a graph execution engine, and obtaining information processing results returned by each of the target algorithm graphs after execution is completed; The information processing results are aggregated to obtain a response result to the information processing request.

2. The method according to claim 1, wherein: In response to the information processing request, a plurality of target algorithm graphs are selected from the algorithm graph library, including: In response to the information processing request, extract at least one key information description field in the information processing request; According to the key information description field, the mapping relationship between the pre-constructed description information and the algorithm graph is queried, and each of the target algorithm graphs is selected in the algorithm graph library.

3. The method according to claim 2, wherein: The key information description field includes: a requester attribute description field and a request scenario description field; The step of querying the mapping relationship between the pre-built description information and the algorithm graph according to the key information description field and selecting each of the target algorithm graphs in the algorithm graph library includes: According to the requester attribute description field, query a first mapping relationship between the requester attribute information and the algorithm graph; Selecting a first type of algorithm graph from the algorithm graph library according to the first mapping relationship; According to the request scene description field, query the second mapping relationship between the request scene information and the algorithm graph; Selecting a second type of algorithm graph from the algorithm graph library according to the second mapping relationship; According to the first type of algorithm graph and the second type of algorithm graph, each of the target algorithm graphs is obtained.

4. The method according to claim 3, wherein: The key information description field also includes: an information processing accuracy description field; The acquiring of each of the target algorithm graphs according to the first type of algorithm graph and the second type of algorithm graph specifically includes: Calling the information processing accuracy determination model to obtain the information processing accuracy of each of the first-type algorithm graphs for the requester attribute description field, and the information processing accuracy of each of the second-type algorithm graphs for the request scenario description field; Among the first type of algorithm diagrams and the second type of algorithm diagrams, select the algorithm diagrams whose information processing accuracy is higher than the information processing accuracy description field; The target algorithm graphs are obtained according to the filtered algorithm graphs.

5. The method according to claim 2, wherein: The key information description field includes: an experiment description field; The step of querying the mapping relationship between the pre-built description information and the algorithm graph according to the key information description field and selecting each of the target algorithm graphs in the algorithm graph library includes: According to the experiment description field, query the third mapping relationship between the experiment information and the algorithm diagram; According to the third mapping relationship, a third type of algorithm graph is selected from the algorithm graph library as each of the target algorithm graphs.

6. The method according to any one of claims 1 to 5, wherein: The concurrent execution of each of the target algorithm graphs by a graph execution engine includes: Respectively obtain the data type of the pre-dependency data required by each target algorithm graph when executing; According to the data type and the information processing request, executing an acquisition operation of all the pre-dependent data required for each of the target algorithm graphs; Constructing a common precondition graph according to the acquired precondition dependency data; After the common pre-graph is used as the common input graph of each of the target algorithm graphs, each of the target algorithm graphs is executed concurrently through a graph execution engine.

7. The method according to claim 6, after the graph execution engine concurrently executes each of the target algorithm graphs, further comprising: After each operator in each of the target algorithm graphs is triggered to execute, the matching processing logic is executed according to the calculation results obtained from the preceding operators of the target algorithm graph to which it belongs and / or the preceding dependency data obtained from the common preceding graph; When it is determined that a target calculation result obtained by executing any operator is relied upon by operators of other algorithm graphs, the target calculation result is updated in real time to the common pre-graph.

8. The method according to claim 7, wherein: After each operator in each of the target algorithm graphs is triggered to execute, the matching processing logic is executed according to the calculation results obtained from the preceding operators of the target algorithm graph to which it belongs and / or the preceding dependency data obtained from the common preceding graph, including: After the target operator is triggered to execute, detecting whether the target operator needs to obtain target predecessor dependency data from the common predecessor graph during execution; When it is determined that the target predecessor dependency data needs to be obtained from the public predecessor graph, detecting whether the target predecessor dependency data is currently stored in the public predecessor graph; If the target pre-dependency data is stored, executing the processing logic matching the target operator according to the target pre-dependency data; If the target pre-dependency data is not stored, then the operation of detecting whether the target pre-dependency data is currently stored in the public pre-dependency graph is returned to be executed at every preset time interval until the target pre-dependency data is detected.

9. The method according to any one of claims 1 to 5, wherein: The summarizing of the information processing results includes: After each of the information processing results is post-processed according to the data processing logic defined in the pre-constructed public post-processing graph, the post-processing results are summarized.

10. The method according to claim 9, wherein: The post-processing of each of the information processing results according to the data processing logic defined in the pre-built public post-processing graph is summarized, specifically including: Counting the total number of information result entries contained in each of the information processing results; If the total number of information result entries exceeds a preset number threshold, the difference between the total number of information result entries and the number threshold is calculated as the number of deleted entries; Obtaining the importance scores of each of the target algorithm graphs respectively, and determining the deletion ratio of the result items of each of the target algorithm graphs according to the importance scores; After sorting the information processing results of each of the target algorithm graphs, deletion and filtering are performed according to the deletion ratio of the matching result items; The deletion and filtering results corresponding to each of the target algorithm graphs are summarized.

11. An information processing device based on a graph execution engine, comprising: An algorithm graph selection module, for selecting a plurality of target algorithm graphs from an algorithm graph library in response to an information processing request; Wherein, an algorithm graph corresponds to an independently developed information processing algorithm, and each algorithm graph contains at least one operator for executing a set processing logic; An algorithm graph concurrent execution module, used to concurrently execute each of the target algorithm graphs through a graph execution engine, and obtain information processing results returned by each of the target algorithm graphs after execution; The result aggregation module is used to aggregate the information processing results to obtain a response result to the information processing request.

12. The device according to claim 11, wherein The algorithm graph selection module comprises: a key information description field extraction unit, configured to extract at least one key information description field in the information processing request in response to the information processing request; A mapping selection unit is used to query the mapping relationship between the pre-constructed description information and the algorithm diagram according to the key information description field, and select each of the target algorithm diagrams in the algorithm diagram library.

13. The device according to claim 12, wherein: The key information description field includes: a requester attribute description field and a request scenario description field; The mapping selection unit includes: A first selection subunit is used to query a first mapping relationship between the requester attribute information and the algorithm graph according to the requester attribute description field; and select a first type of algorithm graph in the algorithm graph library according to the first mapping relationship; A second selection subunit is used to query a second mapping relationship between the request scene information and the algorithm graph according to the request scene description field; and select a second type of algorithm graph in the algorithm graph library according to the second mapping relationship; A result summary subunit is selected to obtain each of the target algorithm graphs based on the first type of algorithm graph and the second type of algorithm graph.

14. The device according to claim 13, wherein: The key information description field also includes: an information processing accuracy description field; The selection result summary subunit is specifically used for: Calling the information processing accuracy determination model to obtain the information processing accuracy of each of the first-type algorithm graphs for the requester attribute description field, and the information processing accuracy of each of the second-type algorithm graphs for the request scenario description field; Among the first type of algorithm diagrams and the second type of algorithm diagrams, select the algorithm diagrams whose information processing accuracy is higher than the information processing accuracy description field; The target algorithm graphs are obtained according to the filtered algorithm graphs.

15. The device according to claim 12, wherein: The key information description field includes: an experiment description field; The mapping selection unit is used to: According to the experiment description field, query the third mapping relationship between the experiment information and the algorithm diagram; According to the third mapping relationship, a third type of algorithm graph is selected from the algorithm graph library as each of the target algorithm graphs.

16. The device according to any one of claims 11 to 15, wherein: The algorithm graph concurrent execution module is used to: Respectively obtain the data type of the pre-dependency data required by each target algorithm graph when executing; According to the data type and the information processing request, executing an acquisition operation of all the pre-dependent data required for each of the target algorithm graphs; Constructing a common precondition graph according to the acquired precondition dependency data; After the common pre-graph is used as the common input graph of each of the target algorithm graphs, each of the target algorithm graphs is executed concurrently through a graph execution engine.

17. The apparatus according to claim 16, further comprising an operator execution module, configured to: After the target algorithm graphs are concurrently executed by the graph execution engine, each operator in each target algorithm graph, after being triggered to execute, executes matching processing logic according to the calculation results obtained from the predecessor operator of the target algorithm graph to which it belongs and / or the predecessor dependency data obtained from the common predecessor graph; When it is determined that a target calculation result obtained by executing any operator is relied upon by operators of other algorithm graphs, the target calculation result is updated in real time to the common pre-graph.

18. The device according to claim 17, wherein: The operator execution module is further used to: After the target operator is triggered to execute, detecting whether the target operator needs to obtain target predecessor dependency data from the common predecessor graph during execution; When it is determined that the target predecessor dependency data needs to be obtained from the public predecessor graph, detecting whether the target predecessor dependency data is currently stored in the public predecessor graph; If the target pre-dependency data is stored, executing the processing logic matching the target operator according to the target pre-dependency data; If the target pre-dependency data is not stored, then the operation of detecting whether the target pre-dependency data is currently stored in the public pre-dependency graph is returned to be executed at every preset time interval until the target pre-dependency data is detected.

19. The device according to any one of claims 11 to 15, wherein: The result summary module is used to: After each of the information processing results is post-processed according to the data processing logic defined in the pre-constructed public post-processing graph, the post-processing results are summarized.

20. The device according to claim 19, wherein The result summary module is further used for: Counting the total number of information result entries contained in each of the information processing results; If the total number of information result entries exceeds a preset number threshold, the difference between the total number of information result entries and the number threshold is calculated as the number of deleted entries; Obtaining the importance scores of each of the target algorithm graphs respectively, and determining the deletion ratio of the result items of each of the target algorithm graphs according to the importance scores; After sorting the information processing results of each of the target algorithm graphs, deletion and filtering are performed according to the deletion ratio of the matching result items; The deletion and filtering results corresponding to each of the target algorithm graphs are summarized.

21. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the information processing method based on the graph execution engine according to any one of claims 1-10.

22. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to enable a computer to execute the information processing method based on a graph execution engine according to any one of claims 1-10.

23. A computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the information processing method based on a graph execution engine as described in any one of claims 1 to 10.

Images